The use of researcher licence in qualitative knowledge production

Lamminpää, S. & C. Rink (toim.) 2014. Demokratia, Demokrati, Democracy, Demokratie. VAKKI-symposiumi XXXIV 13.–14.2.2014. VAKKI Publications 3. Vaasa, (58–72). The use of researcher licence in qualitative knowledge production Riitta Sallinen & Eva Braidwood Languages and Communication University of Oulu Extension School Kvalitatiivisessa tutkimustavassa tutkijan uskottavuus nousee keskeiseen asemaan, koska ihmisen kokemusta tutkivan kvalitatiivisen tutkimuksen aineisto on kielellistä ja tutkija itse tulkitsee ja analysoi aineiston. Tässä tutkimuksessa tarkastelimme tutkimusasetelman vakuuttavuudesta ja tutkijan osoittamasta asiantuntijuudesta syntyvän uskottavuuden, eli ns. ´tutkijalisenssin´ heijastumista kvalitatiivisen tutkimusraportin tekstiin. Oletimme, että tekstilliset vaikutukset liittyisivät väittämien perusteluun. Tutkimusaineistona oli kymmenen psykiatrian alan kvalitatiivista tutkimusartikkelia. Evidentiaalisuus - teoriaan perustuen koodasimme aineiston propositiot niihin liittyvän todisteisuuden mukaan. Oletimme, että ilman argumentointia tai muuta tiedon lähdettä esitettyjen kumottavissa olevien väittämien esiintyminen julkaistuissa tutkimusraporteissa oli osoitusta tutkijalisenssin hyväksytystä käytöstä. Tunnistimme väittämätyyppejä, jotka sisällöltään olivat kumottavia perustelun puuttuessa, mutta usein esiintyivät ilman tiedonlähdevihjettä ja laskimme ko. väittämien esiintyvyyden. Pohdimme myös tutkijalisenssin käyttöä liittyen yhteenveto- ja yleistämisprosesseihin sekä hedging-vihjeiden vaikutusta. Keywords: qualitative research reporting, researcher credibility, evidentiality, claim validation 1 Background Credibility lies at the core of every scientific enquiry. The credibility of research findings stems from documenting the research procedure, the instruments used and the supporting evidence, as well as from the researcher’s familiarity with the field of related studies. Focusing on the study of human experience, qualitative research uses languagebased procedures of knowledge production (Polkinghorne 2007). Such procedures involve generating by means of various discourse-analytical methods scientifically acceptable knowledge claims from participant narrative. As participant narrative is accessible to the readers only in the form of a limited number of quotations presented as evidence for the claims made, the role of the researcher as an interpreting and coordinating instrument of knowledge production becomes seminal. This in turn entails a need for researchers to establish their professional credibility. Researcher credibility is conventionally established by informing the readers of relevant facts in the researcher’s profes- 58 The use of researcher licence in qualitative knowledge production sional, personal, and public background (Polit & Beck 2014: 331). Established researcher credibility is assumed to provide the researchers with a ‘licence’ which justifies their active role in qualitative knowledge production and may also allow some circumvention of the academic tradition of supporting claims with unambiguous and adequate evidence. In this paper we refer to such circumvention of the rules of claim validation as the use of researcher licence (RL). This study investigates the scope of the use of RL acceptable to the community of qualitative researchers. The above research topic arose from the scrutiny of qualitative research articles which constitute one type of text in scientific writing courses for health sciences. Even though the notion of researcher credibility is well established in the literature on qualitative research (Polit & Beck 2014: 331), we did not find analyses in the linguistic literature on how it may be reflected in the textual fabric of qualitative articles. To answer the above question, we studied the instance of statements of findings unsupported by evidence, as well as the instance of other challengeable but unsupported claims in published qualitative psychiatry research papers. These were considered a text-level manifestation of an acceptable degree of the use of RL in qualitative knowledge production. In addition, we considered the effects on the degree of the use of RL of various types of evidentiality markers (Aikhenvald 2004: 63–64) and markers of epistemic modality (e.g., ‘may’ or ‘seem to’ with a hedging effect). We also examined the RL-related effects of linguistic processes such as generalising and summarising, which involve the researcher’s subjective judgment. 2 Corpus and procedure To find out about the frequency and propositional characteristics of unsupported statements of findings and other challengeable but unsupported claims in the target text type, we coded all the propositions of the main text (i.e. excluding the abstract) of ten psychiatry journal articles according to their qualities on a continuum from unchallengeable statements of objective facts or subjective experience to challengeable claims involving subjective judgment. We also double coded the propositions by the evidentiary functions, if any, which they served with respect to adjacent propositions. The total number 59 Riitta Sallinen & Eva Braidwood of propositions analysed in the corpus was 1688. Table 1 below provides citing information and the number of propositions analysed for the articles included in the corpus: Table 1. Information on the corpus Article No. of propositions Source #1 187 BMJ Open2012;2:e000641 #2 201 BMJ 2011;343:d5801 #3 214 BMJ 2010;341:c4184 #4 168 BMJ Open2011;1:e000017 #5 147 BMJ Open2013;3:e002283 #6 164 BMC Psychiatry 2010, 10:8 #7 191 BMC Psychiatry 2013, 13:272 #8 112 Intellect Dev Disabil Jun 2012; 50 (3): 243–250 #9 188 BMJ 2011;343:d5801 # 10 116 J Med Ethics 2011;37:601–605 2.1 Theoretical framework As a theoretical frame for identifying challengeable but unsupported propositions in the corpus, we drew upon the notion of evidentiality. In the literature, evidentiality is defined either in a broader or narrower sense. According to the narrow definition, the term ‘evidentiality’ denotes ‘the ways in which the speaker qualifies a statement by referring to the source of the information’ (Saeed 1997: 133). The difference of opinion among the proponents of the broad and narrow view of evidentiality stems from the issue whether epistemic modality is inherently included in the notion of evidentiality or not. Chafe (1986), for instance, considers evidentiality in the broad sense: “an indication of the source and reliability of a speaker’s knowledge”. A proponent of the narrow definition, de Haan (2005) points out that “[e]videntiality asserts the evidence, while epistemic modality evaluates the evidence” (italics in the original). In this study, our approach was first to find out to what extent knowledge claims in qualitative research articles are qualified by reference to the sources of information, and then to consider the effect of 60 The use of researcher licence in qualitative knowledge production epistemic modality markers on the strength of the evidence provided, and accordingly, on the scope of the researcher licence used. In Table 2 below we provide a comparison of the categories of evidential information described by Aikhenvald (2004: 63–64) and by Chafe (1986: 263). To avoid unnecessary complexity in this report, we adopt the overviewing terms of the first column for further discussion. Table 2. Categories of evidential information Perception-based evidentiality Reasoning-based evidentiality Reporting-based evidentiality 1. VISUAL (SENSORY): information acquired through seeing (Aikhenvald) 3. INFERENCE: conclusion based on visual or tangible evidence (Aikhenvald) (INDUCTION according to Chafe) 5. HEARSAY: reported information without making reference to the person from whom the information was acquired (Aikhenvald) 2. (NON-VISUAL) SENSORY: information acquired through other forms of sensory perception (smell, taste, touch) (Aikhenvald) 4. ASSUMPTION: based on indications other than visible evidence (logic, supposition, general knowledge), with a strong reasoning component (Aikhenvald) (DEDUCTION according to Chafe) 6. QUOTATIVE: reported information with explicit mention of the source (Aikhenvald) The first survey of the corpus suggested that the types of evidentiality used in the target texts are determined by genre conventions (see also Yang 2013). Accordingly, it was necessary to specify the following four genre-typical subtypes of evidentiality: 1) We considered agency by 1st person singular/plural as involving evidentiality based on direct perception (categories 1 and 2 by Aikhenvald). De Haan (2005) also observes that 1st person agency represents the ‘deictic center of evidentiality’. 2) In addition, reference, whether author prominent or subject-matter prominent, was understood to constitute a subcategory of the reporting-based type of evidentiality. 3) We also incorporated in the category of reporting-based evidentiality concurrent self-reference (CSR), i.e. reference to some element of the study whose findings are discussed (“Results indicated that knowing that…” — article #1). Besides references to various elements of the study, the markers of CSR also include the use of past tense (“However, power of attorney for financial decisions… seemed easier” — article #3). CSR constitutes a basic frame for the introduction of research-based knowledge claims (Braidwood & Sallinen 2010). The evidential impact of CSR arises from the fact that a reference to the study involves a 61 Riitta Sallinen & Eva Braidwood reference to the method, data and reasoning-related credibility of the respective study. 4) Moreover, direct quotations from participants’ narratives seemed to serve a function corresponding to that of argument in reasoning-based evidentiality. Contrary to previous studies designed to identify grammaticalised or lexicophraseology-based systems of evidentiality in different languages, we used in the present study the notion of evidentiality as a framework for identifying challengeable but unsupported propositions in the corpus articles. If a proposition did not incorporate any markers referring to perception or reporting as its source of knowledge and was not supported by reasoning, it was considered to be challengeable but unsupported. Propositions including adequate (see 3.2.1 for variation in the evidentiary force of different types of evidentiality) textual or contextual markers of evidentiality as well as those stating subjective experience were regarded as unchallengeable. Our study was based on the premise that, by the rule of academic knowledge production, propositions challengeable due to a lack of perception or reporting-based evidentiality need to be supported by reasoning-based evidentiality to gain a status as valid statements of fact. Exceptions to this rule were considered as representing the use of RL. Still another difference in comparison with earlier studies dealing with evidentiality was that we applied the notion of evidentiality to the description of the sources of knowledge claims on both sentence- and text-level. We did so because the inductive and deductive processes of reasoning, which typically constitute sources of knowledge in scientific articles, are frequently manifested in wider text-structural patterns. 2.2 Preliminary coding categories To identify and describe in a relevant way exceptions to the academic rule of claim validation, we needed a system of categorization which 1) allowed for the description of the qualities of propositions on the continuum from unchallengeability to challengeability, 2) provided a representation of the patterns of evidentiality based on the premise that challengeable claims need to be validated, and 3) indicated the genre-related functions of target propositions. As pointed out above, the evaluation of the challengeability 62 The use of researcher licence in qualitative knowledge production or validity of propositions was grounded on the theory of evidentiality. The theory also provided the means for describing various patterns of evidentiality. The description of the categories of genre-typical propositions, on the contrary, arose from a scrutiny of the corpus articles and our familiarity with the genre of qualitative research articles (see also Sallinen & Braidwood 2012). Because the categories defined in this study for the validity of propositions as factual statements are genre-specific, they do not claim any universal representativeness. The preliminary categories within which we looked for the exceptional cases of challengeable but unsupported propositions included the following: Table 3. Preliminary code categories Statements Conclusions articulating facts evidentiality through first-person agency / implied source of verification / as canonized knowledge findings-external claims * ** findings statements drawing associations (e.g. analogical, logical, categorical, etc.) between objects, notions, claims and phenomena * ** based on preceding evidence, unchallengeable Reportive statements of participant narrative (RPN) incorporate reportative evidentiality marker unchallengeable but frequently supported by reportative evidence Descriptive statements of participant experience (DPE) / behaviour (DPB) Interpretations of meaning of PN reportative evidentiality marker missing * ** unchallengeable as a conventionalized researcher function in qualitative study *** challengeable if no reasoning follows * unchallengeable, valid for the speaker Claims Inferences Evaluations of value of study Expressives *challengeable if no evidence provided **supportable by reasoning-based or reportative evidence ***may be supported by reasoning-based evidence The closer analysis, which was focused on identifying, describing, and determining the incidence of unsupported propositions in the above categories of challengeable propositions produced further subcategories for some of these main categories. 63 Riitta Sallinen & Eva Braidwood 3 Findings The description of the types and frequency of propositions with challengeable validity but no or inadequate supporting evidence in the corpus articles was considered to represent a description of the scope of the researcher licence used by the authors of the articles in the corpus. Corpus-based categories for challengeable but unsupported propositions: Table 4 below provides a description of the emergent categories of challengeable but unsupported propositions by reference to their criteria for challengeability, their frequency and location in the qualitative research articles of the corpus and their function in the genre-determined process of research reporting. In addition, their unsupported incidence is explained in terms of observed variation in the function of RL use. A closer observation of the categories introduced in Table 4 suggests two different functions in the use of RL. Unsupported propositions such as claims of common knowledge in the field, niche claims and evaluations of the value of the study seem to represent 1) genrespecific conventionalized use of RL; while concluding inferences, unsupported comparative, explanatory, consequential and implicational inferences, and DPE and DPB with no supporting quotation constitute 2) findings-related use of RL. The former type seems to represent a constant in the volume of RL use and is more closely related to genre rhetoric than research. The latter use of RL, on the contrary, is associated with knowledge production in the Results and Discussion sections, and accordingly constitutes the focus of this study. 64 The use of researcher licence in qualitative knowledge production Table 4. Corpus-based categories for challengeable but unsupported propositions Category Example Criteria for challengeability Location & frequency Genrebased function Explanation for unsupported incidence & comment Unsupported COCK represent genre conventions -based use of RL. Other types of findings-external claims are usually supported by reference. Unsupported CON represent genre conventions -based use of RL. Even though CON is assumed to be based on literature review, references to supporting evidence are frequently omitted. Unsupported CONCINF represent findings-related use of RL. Claims of common knowledge in the field (COCK) (1) Prevention of suicide is a major public health concern in both developed and developing nations. (article #2) Lack of evidentiality markers. Introduction: 3.6% (9/253) Findingsexternal knowledge claim describing the background of the study Claims of niche (CON) (2) However, limited evidence exists on which to build action plans. (article #2) Lack of evidentiality markers. If the proposition incorporates a phrase such as ‘To our knowledge…’ with a reference to 1st person perception, the claim is not challengeable. Introduction: 9 in the 10 articles of the corpus (3) In part, taking over their (finances) management represented a milestone in deterioration and role change and … a role reversal.´ (article #3) Lack of perception and reasoning – based evidentiality markers. May incorporate CSR as a weak marker of evidentiality. Results: 6.7% (42/621)* Discussion: 11% (50/445)* Findings statements are often repeated without argumentative support in the Discussion due to the summarizing function of the section. Such repetitive propositions were excluded from the figure for the Discussion section. Results: 1% (7/621)* Discussion: 3% (14/445)* Findingsexternal knowledge claim concerning the existence, adequacy and appropriateness of previous research findings. Findings statement communicating inference-based findings Concluding inferences based on research data inaccessible to readers (CONCINF)) (4) Unless doctors are able to acknowledge realistic limitations and vulnerability, they will need to use denial to protect themselves from conscious awareness of their vulnerability and fallibility, and minimise these to others. (article #4) Comparative inferences (COMPINF)) (5) Those with South-Asian values seem to be more depressed and … (article #1) Lack of perception- and reasoning -based evidentiality markers. Comparative inferences may incorporate CSR as a weak marker of evidentiality. Findings statement communicating inference-based comparisons This highlights the role of the researcher as an instrument of knowledge production. Unsupported COMPINF represent findings-related use of RL. Comparative inferences are separate from comparisons of present and previous research findings, which are usually supported by reference. 65 Riitta Sallinen & Eva Braidwood (Respect for the person’s privacy and autonomy prevented some from confronting problems directly.) (6) Trust in the person’s capacity to resolve his or her own problems may have played a part in this. (article #2) Implicational (7) It may be that improving the inferences quality and effec(IMPINF) tiveness of informal doctor-to-doctor conversations is an area where a difference can be made in improving doctors’ access to mentalhealth services. (article #5) Evaluations of (8) "This is an inthe value of the depth study of an information-rich study (EOV) group in an area where more understanding is acutely needed." (article #4) Lack of evidentiality markers. Challengeability is frequently alleviated by the use of epistemic modality markers for hedging. Results: 4% (25/621)* Discussion: 4,5% (20/445)* Elaborative inference from findings which states their causes and consequences Unsupported E&CINF represent findings-related use of RL. Lack of evidentiality markers. Challengeability is frequently alleviated by the use of epistemic modality markers for hedging. Discussion: 9% (42/445)* Elaborative inference from findings which states their practical implications Unsupported IMPINF represent findings-related use of RL. Lack of evidentiality markers. Discussion: 16 in the 10 articles of the corpus Positive evaluation of the value of the study by the researcher(s) Descriptions of participant experience (DPE)/ behavior (DPB) Lack of evidentiality Results: 8% markers other than CSR. (87/1066) of the findings statements in the results sections of the corpus articles Unsupported EOV represent genre conventions-based use of RL. Even though this proposition conveys a multidimensional positive judgment of the research setting, the researcher conventionally present it without supporting it by argument. Unsupported DPE & DPB represent findings-related use of RL. As an alternative to reporting participant narrative (RPN), researchers seem to revert to describing participant experience or participant behavior. This involves omitting the marker of reporting-based evidentiality. Explanatory and consequential inferences (E&CI) (9) Going to see the general practitioner together helped, for example, as did the doctor writing to the patient. (article #3) Findings statement describing PE or PB on the basis of information gained through participant narratives (PN). *of all propositions in the respective section To sum up, we propose on the basis of the propositional analysis of the corpus articles that concluding, comparative, explanatory & consequential, and implicational infer- 66 The use of researcher licence in qualitative knowledge production ences, as well as DPE and DPB as defined in Table 3 constitute the core of the propositions in which the use of RL is manifested.1 As one of the aims of this study was also to define the scope of RL use, we conducted a small-scale quantitative analysis of the data. Table 5 below illustrates the volume of the findings-related use of RL in our corpus by displaying the frequency of challengeable but unsupported propositions in the Results and Discussion sections of the corpus articles. The higher percentage of unsupported propositions in the Discussion section is most likely due to the causal and implicational comments which pertain to the deliberation of the findings. The means for the frequencies in the two sections suggest a relatively high incidence in the corpus of statements of findings or other findings-related propositions, which from the point of view of the rules of academic knowledge production are exceptional. However, the dispersion of the frequency values from the mean also seems to be high. Based on the notion of standard deviation, 68.27 % of the measured values in normally distributed data exist at the distance of one standard deviation from the mean. In our case this rule seems to allow the prediction that in 68.27 % of qualitative research reports, the percentage of challengeable but unsupported propositions varies between 15 % and 35 % in the Results section and between 21 % and 43 % in the Discussion section. The small size of the corpus may, however, affect the reliability of such prediction. Yet, it seems to be obvious that the notable deviation from the mean incidence of challengeable but unsupported propositions in the corpus articles refers to individual differences in the degree to which researchers rely on the use of RL. A further explanatory factor for the variation may be differences in the extent to which different journals accept the use of RL. 1 Although the findings of this study highlight the unsupported occurrence of these findings-related propositions in the Results and Discussion sections of qualitative research articles, they may, however, also incorporate supporting evidence as any inferential proposition. 67 Riitta Sallinen & Eva Braidwood Table 5. The frequency of challengeable but unsupported propositions in the Results and Discussion sections of the corpus articles Article 1 2 3 4 5 6 7 8 9 10 Mean Standard deviation Results 12/63 =19% 17/101 =17% 25/71 =35% 6/48 =13% 18/56 =32% 17/56 =30% 20/95 =21% 11/29 =37% 9/27 =33% 6/75 =8% 25% 10.16 Discussion 17/55 =31% 12/53 =23% 31/73 =43% 9/43 =21% 16/40 =40% 4/33 =12% 13/38 =34% 11/31 =35% 30/59 =50% 5/19 =26% 32% 11.35 Further observations on the use of researcher licence (RL) in qualitative research articles: The idea given by Table 5 of the extent to which RL is used in qualitative psychiatry research articles may however be only tentative. This is because the degree of reliance on RL is most likely affected by co-occurring features such as variation in the validating force of different forms of evidentiality, the amount of subjective judgment involved in the functions performed by the researcher as an instrument of knowledge production and the use of markers of epistemic modality. We will discuss the influence of such cooccurring phenomena in the following sections. Use of RL affected by the epistemic momentum of different forms of evidentiality: The extent of the use of RL in qualitative research articles is likely to be affected by the types of evidentiality distinctive of the target discipline. While reference and reasoning, which draw on established knowledge and inferential logic, are seen as authorized means of academic knowledge production, the use of reportative evidentiality seems to reflect lower stability of epistemic momentum. This is the case particularly in propositions reporting participant narrative (RPN), (e.g. “Many informants suggested that the deceased did not give out clear distress signals” – article #3), which may or may not be followed by supporting quotation. In our corpus 46 % (129/276) of the RPNs were followed by a supporting quotation. If the epistemic momentum of reportative evidentiality of the above type is questionable, its unsupported instances can be assumed to draw on the use of RL. In addition, another type of reportative evidentiality with difficult-todefine epistemic power seems to be concurrent self-reference (CSR), (e.g. “Our data suggest that …– article #2). Even though CSR refers to the complete set of research 68 The use of researcher licence in qualitative knowledge production tools for validity, it cannot fully guarantee the validity of the knowledge claims made at this stage of the process of knowledge production where the researchers are proposing the study for consideration to the academic community. It seems obvious, however, that incorporating a frame of CSR reduces reliance on RL. Use of RL arising from the functions of the researcher as an instrument of knowledge production — Summaries of participant narrative: Even though the ideographic nature of knowledge is highlighted in the constructivist conception of knowledge (von Glaserfeld 1996), social constructivists consider consensus between different subjects ‘the ultimate criterion to judge knowledge’ (Heylighen 1993). The latter position seems to be reflected in the fact that summaries of participant narrative represent an important outcome of a qualitative researcher’s activity as an instrument of knowledge production. In our corpus, 74 % (242/324) of the propositions reporting participant narrative had a plural subject (Participants said/reported/described that …). The construction of such summaries of spoken language involves cognitive processes, such as interpretation and comparison, to recognize commonalities in the narratives of the participants, as well as an element of generalisation because similarity is still different from sameness. These processes necessarily incorporate subjective judgment, which means that the researchers are using their RL (e.g.,“The participants in the study related a complex, often distressing journey in negotiating an appropriate mixture of care and control within the care system” – article #3). Unargued generalisation: The use of RL also seems to be highlighted in generalisations of various types not grounded on argument. Even though generalisation is not a means of knowledge production that is inherently in line with the constructivist epistemology (Lincoln & Guba 1985: 110), generalisations with more restricted coverage and supporting argument are proposed by contemporary theorists (Mayring 2007). Occasionally, however, qualitative researchers seem to venture generalizing from participant experience to target population experience with no supporting argument. At the level of the text this shows in the absence of the frame of concurrent self-reference and in the use of the present tense. This is the case in Example 1, which shows a generalisation from the finding stated in 69 Riitta Sallinen & Eva Braidwood the preceding proposition originally presented in the Results section. Generalisations are also frequently manifested in abstracted findings statements transcribed into the language of the discipline (Example 2). (1) (2) [Participants consistently described difficulties with the responsibility of making a decision for another adult, denial and resistance by the person with memory problems, and barriers to accessing services. (article #2, Results)] →→ Difficulties in decision making for people without capacity are often aggravated by their active resistance. (article #2, Discussion) The study findings highlight the multi-dimensional nature of the aetiological models of depression among South Asian immigrant women in Toronto, Canada. (article #1) Epistemic modality: Hedging, by which “writers can present a proposition as an opinion rather than a fact” (Hyland 1998), is a pervasive type of epistemic modality in research articles and it seems to counteract the use of RL. This is because the burden of evidence is reduced for a hedged claim. In our corpus 22 % (21/92) of concluding inferences, 33 % (17/51) of unsupported explanatory and consequential inferences, and 25 % (10/40) of unsupported implicational inferences incorporated a hedging epistemic modality marker. 4 Conclusion This study investigated the textual manifestations of the use of researcher licence in qualitative psychiatry research reports. It demonstrated that qualitative research articles reporting studies based on the use of narrative data typically incorporate a number of challengeable but unsupported propositions. Propositions such as claims of common knowledge in the field, claims of niche, and evaluations of the value of the study, which often appear unsupported, serve genre-specific functions and seem to be conventionally acceptable. The idea of the researcher operating on a licence approved by the respective research community seems to be manifested, however, most clearly in concluding inferences and DPE and DPB which communicate research findings based on data inaccessible to the readers. Such practice may have gained acceptability because qualitative study as a language-based process cannot depict complete chains of reasoning from data to final conclusions. The unsupported incidence of other findings-related types of challengeable propositions, i.e., comparative, explanatory & consequential, and implication- 70 The use of researcher licence in qualitative knowledge production al inferences, on the contrary, seems to be less contradictory to the rule of claim validation due to the inherent element of subjectivity involved in such inferences. The statistical analysis of the incidence of challengeable but unsupported findings-related propositions in the corpus texts suggested a frequency between 15 % – 35 % of such propositions in the Results section, and one between 21 % – 43 % in the Discussion section for 68 % of qualitative research reports. Even though such generalisation may be risky on the basis of our small corpus, it seems that a notable number of findings-related propositions are affected by the use of researcher licence i.e., the researcher’s subjective judgment. On the other hand, the above figures for standard deviation also indicate that there is a great deal of individual variation between researchers. The scope of the use of RL is, however, defined not only by the frequency of the various types of challengeable but locally unsupported propositions. It is also affected by the subjective elements embedded in the processing of the data with the researcher as the analysing, interpreting and generalising intelligence, the epistemic momentum of the types of evidentiality used, and the use of markers of epistemic modality. We presumed that the scope of the use of researcher licence identified in this study is acceptable to the academic community since the corpus consisted of articles published in peer-reviewed journals. However, the acceptability of the use of a researcher licence also seems to be embedded in the social constructivist epistemology according to which knowledge arises from the consensus of multiple participants (Creswell 2003) in conceiving experience in a social and historical context. References Aikhenvald, Alexandra (2004). Evidentiality. Oxford, UK: Oxford University Press. Braidwood, Eva & Riitta Sallinen (2010). Rhetorical Patterns of Claim Validation in Qualitative and Quantitative Research Reporting. In: Language and Emotions. VAKKI-juhlasymposiumi XXX, 298–310. Eds Niina Nissilä & Nestori Siponkoski. Vaasa: Vaasan yliopisto. Chafe, Wallace (1986). Evidentiality in English Conversation and Academic Writing. In: Evidentiality: the linguistic coding of epistemology. Advances in discourse processes, vol. XX, 261–272. Eds Wallace Chafe & Johanna Nichols. New Jersey: Ablex Publishing Corporation. 71 Riitta Sallinen & Eva Braidwood Creswell, John W. (2003). Research Design: Qualitative, Quantitative, and Mixed Method Approaches. 2nd edition. Thousand Oaks, California: Sage Publications. de Haan, Ferdinand (2005). Encoding Speaker Perspective: evidentials. In: Linguistic diversity and language theory, 377–396. Eds Zygmunt Frajzyngier, Adam Hodges & David Roods. Amsterdam: Benjamins. Heylighen, Francis (1993). Selection Criteria for the Evolution of Knowledge. [online]. In: Proc. 13th Int. Congress on Cybernetics (Association Internat. de Cybern_(c)tique, Namur), 524–528. [cited 23.4.2014]. Available at: http://pespmc1.vub.ac.be/papers/selcriteriaknow.pdf Hyland, Ken (1998). Hedging in Scientific Research Articles. Philadelphia: John Benjamins. Lincoln, Yvonne & Egon Guba (1985). Naturalistic inquiry. Beverly Hills: Sage Publ. Mayring, Phillip (2007). On Generalisation in Qualitatively Oriented Research. [online]. Forum: Qualitative Social Research. [cited 24.1.2014]. Available at: http://www.qualitative-research.net/index.php/fqs/article/view/291/641 Polit, Denise F. & Cheryl Beck (2014). Essentials of nursing research: Appraising evidence for nursing practice. 8th edition. Philadelphia: Lippincott Williams & Wilkins. Polkinghorne, Donald (2007). Validity issues in narrative research. Qualitative Inquiry 13 (4), 471–486. Saeed, John I. (1997). Semantics. Oxford: Blackwell. Sallinen, Riitta & Eva Braidwood (2012). From Participant Narrative to scientific knowledge: Linguistic processing of qualitative data in nursing science research reporting. In: Languages in Motion, VAKKI-symposiumi XXXII, 298–310. Eds Niina Nissilä & Nestori Siponkoski. Vaasa: Vaasan yliopisto. von Glaserfeld, Ernst (1996). Introduction: Aspects of constructivism. In: Constructivism: Theory, perspectives and practice. Ed. Catherine Fosnot. New York: teachers College Press. Yang, Linxiu (2013). Evaluative Functions of reporting Evidentials in English Research Articles of Applied Linguistics. Open Journal of Modern Linguistics 3.2, 119–126. 72